From Foresight, Vol. 7, No. 4
published 2001
The number of Americans making online purchases in 2000 jumped to 34 percent from 24 percent in 1999, according to a survey by the Conference Board and NFO Research.(1) Nearly 80 percent of Internet users shopped online for gifts during the 2000 holiday season, up from about 70 percent in 1999 according to PricewaterhouseCoopers.(2) Moreover, 74 percent of online shoppers actually purchased gifts online during the 2000 holiday season, compared to 67 percent in 1999.(3) According to eMarketer, the average U.S. Internet buyer purchased $627 worth of goods online in 2000, an increase from $500 in 1999.(4) As a result of more Americans spending more money online, total consumer online shopping revenues for 2000 are estimated by various sources to range from $23.1 billion to $109 billion (see Figure 1). By comparison, in 1999 these estimates ranged from $3.9 billion to $36 billion.
Figure 1: Comparative Estimates: U.S. Consumer Online Shopping Revenues for 2000
We present data in this article about online buyers in Kentucky and explore factors that may help explain the probability of knowledge of the use tax and willingness to pay it. Consumers are obligated to pay sales or use tax on the products they purchase over the Internet, but it is widely believed that few do. Therefore, the number of Kentuckians making online purchases has implications for the state budget as revenue is lost due to unpaid use taxes.
In this section we examine Internet access and usage, a necessary precondition for making an online purchase. A number of studies by the U.S. Department of Commerce’s National Telecommunications & Information Administration (NTIA) have found that Americans are becoming increasingly interconnected in the wired community.(5) As illustrated in Figure 2, the percentage of U.S. households with computers and Internet access increased rapidly from 1994 to 2000.(6)
Figure 2: Percent of U.S. Households with a Computer and Internet Access
Kentucky is competitive with neighboring states with respect to the percentage of households with Internet access (Figure 3). According to Census estimates, 37 percent of Kentucky households used the Internet in 2000, compared to about 42 percent nationally.
Figure 3: Estimated Percent of Households Using the Internet in 2000, Selected States
Three statewide surveys conducted for the Kentucky Long-Term Policy Research Center by the University of Kentucky Survey Research Center find that, like most Americans, citizens of the Commonwealth are moving rapidly into the Information Age.(7) In 1996, we found that about 26 percent of adults in Kentucky had used the Internet. In 1998, rates of Internet access had increased to 42 percent. And in 2000 we find that an estimated 63 percent of surveyed Kentucky adults have accessed the Internet in the past year.(8) Five years ago, we found that younger people and more educated people were the most likely to have used the Internet. While that remains true today, Internet use has soared at all age and education levels. Moreover, regional disparities in Internet use are beginning to narrow.(9)
Currently, the data suggest that Kentuckians are accessing the Internet at rates just below the national average. However, as Internet access and usage in Kentucky increase, Kentuckians will likely become more fully integrated into the digital economy. In the sections below, we present survey data about online purchasers in Kentucky and explore the factors that may help explain the probability of knowledge of the use tax and willingness to pay it.
An estimated 32 percent of Kentucky adults said “Yes” when asked if they have ever made a purchase over the Internet.(10) This is just below the U.S. estimate of 34 percent.(11) This represents a remarkable increase in the number of Kentuckians making online purchases. In 1999 we estimated the percentage about 18 percent.(12)
Who are these cybershoppers? In 1999 we reported that a Kentuckian shopping on the web is most likely a college-educated male, earning at least $50,000 annually, and living in the urban triangle.(13) Table 1 shows how rapidly online buying has permeated virtually all socioeconomic and demographic categories. For example, the 2000 data show that women are just as likely as men to make online purchases; urban triangle residents are still more likely to make online purchases, but the other regions are rapidly gaining ground; and almost all education, income, and age categories experienced a double digit increase in the percentage of individuals who indicated they have made an online purchase.
Table 1: Have you ever purchased a product over the Internet?
Because consumers are expected to spend more and more money online in the future, our understanding of who knows about the use tax and who pays it may become increasingly important to state budgets. The following questions were asked on the University of Kentucky Survey Research Center’s spring 1999 and fall 2000 surveys:
If you make Internet or catalogue purchases from out-of-state companies, Kentucky law requires that you pay Kentucky sales tax. If the company does not collect Kentucky sales tax, it is your responsibility to report the purchase on your state income tax form and pay the 6 percent tax. This is called the use tax. Have you heard of this law? (Yes or No)
If you were to make Internet or catalogue purchases from out-of-state companies that did not collect Kentucky sales tax, how likely is it that you would report those purchases on your state income tax form? (Very Likely, Somewhat Likely, Somewhat Unlikely, Very Unlikely).
In 1999 we reported that only 37 percent of Kentuckians say that they have heard of the use tax,(14) and by fall 2000 the percentage had inched up to around 40 percent,(15) not a statistically significant difference. And when asked in the 2000 survey to give the likelihood of actually reporting the use tax, we find that 32 percent said “very likely,” 20 percent replied “somewhat likely,” 14 percent said “somewhat unlikely,” 31 percent replied “very unlikely,” and 3 percent said that they did not file state income tax.(16)
These survey results show that about 52 percent indicated that they are either very or somewhat likely to pay the use tax. However, these individuals are probably overly optimistic about their likelihood of paying the use tax. A study by the U.S. Advisory Commission on Intergovernmental Relations, for example, estimates that 16.5 percent of use taxes are actually paid, and a report by Ernst & Young assumes that only “4 percent of taxable business-to-consumer e-commerce sales result in sales and use tax payments.”(17)
Nevertheless, it would seem that additional efforts to educate the public about the use tax might result in a higher compliance rate. First, six out of ten Kentucky adults do not know about the requirement to pay the use tax. Obviously, individuals won’t pay a tax that they do not know about. Second, as Figure 4 illustrates, individuals who say they have heard about the use tax are more likely to say they will pay it. For example, of those individuals who say they are very likely to pay the use tax, 37 percent have also heard of the use tax, where 30 percent have not. Conversely, those who say that they are very unlikely to pay the use tax have less familiarity with the tax.
Figure 4: Likelihood One Would Pay the Use Tax by Whether One Had Previously Heard of It
We used a statistical model to estimate the effect of a series of variables on the probability that an individual would report being very or somewhat likely to pay the use tax. For example, are individuals in urban Kentucky more or less likely than those living in rural areas to pay the use tax? Are individuals in households with higher income levels more or less likely to indicate a willingness to pay the tax? And what effect do education, gender, and age have on their expressed willingness to comply with the use tax law? This kind of analysis enables us to estimate the effect of any one variable, like income level, on the probability of an individual’s expressed willingness to comply with the use tax requirement, while holding all other variables constant. For example, the model enables us to take two individuals living in an urban area with the same gender, age, and education and estimate the effect of income level on the probability of complying with the use tax law.(18)
Figure 5 illustrates two of the more interesting results of the statistical analysis. First, individuals who had previously heard of the use tax have a higher probability of reporting that they are either very or somewhat likely to pay the use tax (while holding all other variables constant).(19) Second, as an individual’s income increases, the probability they would indicate a willingness to comply with the use tax law decreases.(20)
Internet commerce has received so much attention from the media and from policymakers that many people probably believe the Internet has surpassed more traditional forms of mail-order purchases, but that is not the case. According to the National Mail Order Association, U.S. consumer mail order sales were worth $185 billion in 1998, with Internet sales accounting for about $5.6 billion, about 3 percent of the total for consumer mail order sales.(21) Indeed, it is conventional wisdom that the bulk of lost revenue currently is due not to Internet sales but to mail-order sales. However, concern in Kentucky and across the country about lost sales and use tax revenue is not prompted by the size of the problem today, but rather by the potential size of the problem a few years from now.
Estimates vary considerably, but they all suggest that the value of Internet sales will be considerably higher in a few years (see Table 2). For example, according to IDC, a technology market analysis company, online (business-to-consumer) sales in the U.S. are expected to generate over $75 billion by 2003 compared to expected revenue of $24 billion in 1999.(22) And this estimate is on the low end (but not the lowest) compared to what other market research firms have forecasted. Another technology market research firm, eMarketer, has forecasted that “revenues from online consumer retail shopping are projected to grow to $125.6 billion by 2004, representing an increase of 240 percent from the $37 billion projected by year-end 2000.”(23) This projection represents a “middle of the road” assessment. On the high end (but certainly not the highest of available forecasts), the Giga Information Group, an e-business advisor, projects that “United States business-to-consumer (B2C) sales over the Internet will grow from an estimated $25 billion in 1999 to $152 billion in 2002 and extend that forecast to $233 billion in 2004.”(24) As a result of these kinds of forecasts, there is widespread belief, if not fear, that while lost revenue may constitute a small percentage of the general fund today, it could be considerably larger in the future. We have estimated the amount of lost sales and use tax from consumer Internet purchases by generating three alternative scenarios (low, moderate, and high-use tax losses) that take into account the range of factors listed below:
Projected U.S. B2C Internet sales. We used the three projections listed in Table 2.
Kentucky’s lower retail consumer expenditures (relative to the U.S. average). Retail sales per household in Kentucky were 92.6 percent of the national average in 1997.(25)
Kentucky’s lower Internet purchasing rate (compared to the U.S. average). An estimated 32 percent of Kentucky adults have made an Internet purchase, compared to 34 percent of Americans.
Substitution effect. A lot of the business-to-consumer Internet sales should not be counted as new sales because they would have taken place as telephone or catalog sales in the absence of the Internet. A study conducted by Ernst & Young assumes that 20 percent of online sales would have taken place in the absence of the Internet and they caution that this is probably a low estimate.(26) In our “high” scenario then, we count only 80 percent of projected sales. We decided to double the percentage in our “low” estimate and use the average of the two for our “moderate” estimate.
Taxability. Not all products and services purchased online are taxable. Our “high” scenario estimate of 42 percent comes from Goolsbee, an economist at the University of Chicago.(27) The “low” scenario estimate from the Ernst & Young report(28) and the “moderate” scenario estimate is the midpoint between the other two estimates.
Use Tax Consumer Compliance Rate. Estimates vary on the percentage of use tax that is remitted to states. In the “high” scenario we assume a 4 percent compliance rate,(29) in the “moderate” scenario we assume a 16.5 percent compliance rate,(30) and in the “low” scenario we assume a 52 percent compliance rate.(31)
Table 2: Projected U.S. Business-to-Consumer Internet Sales
We present a summary of our assumptions in generating the scenarios in Table 3.(32)
Figure 6 shows the estimated lost sales and use tax from Internet consumer purchases from 1999 to 2004 based on three scenarios that reflect a range of estimates for the factors listed in Table 3. Projected annual losses range from $7.6 million to $57 million by 2004. And these amounts range from 0.3 percent to 2 percent of Kentucky’s projected sales tax revenue in 2004. While the percentages seem small, the actual dollar losses are not. For example, the estimated cumulative losses from 1999 to 2004 are $27 million for the low estimate, $74 million for the middle estimate, and $187 million for the high.
Figure 6: Estimates of Lost Sales and Use Tax Revenue from Consumer Internet Purchases in KY
Kentucky consumers will likely be making an increasing percentage of their expenditures on the Internet in the future. This is evidenced by the increase in online shoppers from 18 percent in spring 1999 to 32 percent in fall 2000. Yet, the percentage of Kentuckians who indicated some awareness of the use tax remained essentially unchanged over the same time period. In 1999 an estimated 37 percent of Kentucky adults said they had heard of the use tax and by 2000 the estimate was 40 percent—a statistically insignificant difference.
As more Kentuckians shop online, policymakers will be challenged to develop new and better ways to increase the use tax compliance rate. While these survey results show that over half of Kentuckians say that they are very or somewhat likely to pay their use tax, other research suggests that no more than 16 or 17 percent of the use tax owed nationally is actually paid.(33)
Educating consumers about the use tax could be a good first step. A significant percentage, about 60 percent, does not even know about the requirement to pay the use tax. And our analysis shows that consumers who have heard about the use tax have a much higher probability of indicating they are either very or somewhat likely to comply with the use tax requirement.
The growth of online commerce could have serious implications for the state budget, resulting in several million dollars of unpaid use tax. However, the Streamlined Sales Tax Project and the Uniform Sales and Use Tax Administration Act represent attempts to simplify sales and use tax collection laws across the country and will likely lead to increased collection of these taxes by retailers.(34) Nonetheless, some analysts have cautioned that Internet-based tax-compliance systems will not be in broad use for years.
Consequently, if online purchases continue to increase while use tax compliance remains low, the sales and use tax base will gradually erode. Other sources of revenue will have to be found or state expenditures will have to be adjusted accordingly. Also, the use tax is regressive, since individuals with higher incomes appear to be more likely to shop online and less likely to comply with the use tax requirement. Thus, the results presented in this article suggest that tax reform might be in Kentucky’s future.
* Mr. Childress is the Executive Director of the Kentucky Long-Term Policy Research Center. Return to text.
1. Lori Enos, “Report: More U.S. Households Shopping Online,”
E-Commerce Times 9 Nov. 2000, 5 Jan. 2001 http://www.ecommercetimes.com/news/articles2000/001109-4.shtml Return to text.
3. “Holiday 2000 ….” Return to text.
4. “New eCommerce B2C Report: Consumer eCommerce Sales Will Reach $37 Billion by Year-End 2000, to Top $100 Billion by 2003,” Business Wire, Inc. 26 April 2000, 5 Jan. 2001 http://www.nexis.com. Return to text.
5. National Telecommunications and Information Administration (NTIA), Falling Through The Net: Defining the Digital Divide (U.S. Department of Commerce, Washington, D.C., 1999) NTIA Web site, 2 Dec. 1999 http://www.ntia.doc.gov/ntiahome/fttn99/contents.html. The 2000 data are from Falling Through the Net. Return to text.
6. The data presented from the NTIA are household-level data. The data presented from the University of Kentucky Survey Research Center are individual-level data. Consequently, the two sources of data cannot be compared. Return to text.
7. Michal Smith-Mello, Michael T. Childress, Amy Watts, and John F. Watkins, Challenges for the New Century (Frankfort: Kentucky Long-Term Policy Research Center, 2000) 53-65. These estimates are based on survey data collected by the University of Kentucky Survey Research Center. Return to text.
8. Smith-Mello, Childress, Watts, and Watkins 60. Return to text.
9. Smith-Mello, Childress, Watts, and Watkins. Return to text.
10. The survey asked, “Have you ever purchased any products or services from any organization online through the World Wide Web? [If yes, was it products, services, or both?].” The survey was conducted by the University of Kentucky Survey Research Center. Households were selected using random-digit dialing, a procedure giving every residential telephone line in Kentucky an equal probability of being called. Calls were made from October 28 until November 21, 2000. The sample includes noninstitutionalized Kentuckians 18 years of age or older. There were 859 completed interviews. The margin of error is approximately ±3.3 percentage points at the 95 percent confidence level. Return to text.
11. Enos. Return to text.
12. Michael T. Childress, Robert W. Cox, Merl M. Hackbart, Charles W. Martie, Kevin O’Neil and Peter Schirmer, Collecting Taxes in the Cyberage (Frankfort: Kentucky Long-Term Policy Research Center, 1999) 13. Return to text.
13. Using a logistic regression model, we find statistically significant relationships between the probability of buying something online and education (alpha = .05), income (.05), gender (.10), age (.05), and urban location (.05). Return to text.
14. Sample size = 630. Return to text.
15. Sample size = 850. Return to text.
16. Sample size = 819. Return to text.
17. Robert J. Cline and Thomas S. Neubig, “The Sky Is Not Falling: Why State and Local Revenues Were Not Significantly Impacted by the Internet in 1998,” (Ernst & Young Economics Consulting and Quantitative Analysis, 1999) 7. The U.S. Advisory Commission on Intergovernmental Relations reports a higher rate of 16.5 percent in Taxation of Interstate Mail Order Sales: 1994 Revenue Estimates. Return to text.
18. We used a cumulative logit model for ordinal responses. Technical information about the model is available from the author by request. Return to text.
19. Chi-Square is equal to 17.9 (Pr > ChiSq = <.0001). Return to text.
20. Chi-Square is equal to 3.15 (Pr > ChiSq = .0761). Return to text.
21. National Mail Order Association, “1998 Mail Order Sales Results,” 31 Jan. 2001 <http://www.nmoa.com/Library/1998sale.htm>. The National Mail Order Association includes Internet sales as mail order sales. Return to text.
22. iWord, “World Wide Web: The Global Internet Opportunity Unfolds,” March 2000, 30 Jan. 2001 <http://www.iword.com/iword51/iword51.html>. Return to text.
23. “New eCommerce B2C Report Reveals Consumer e-Commerce Sales Will Reach $126 Billion by 2004, an Increase of 240% from $37 Billion in 2000,” Business Wire, Inc. 31 Oct. 2000, 5 Jan. 2001 <http://www.nexis.com>. Return to text.
24. “Giga Information Group Predicts Strong B2C E-Commerce Growth Despite Dot Com Shakeout,” News Release 18 July 2000, 30 Jan. 2001 <http://www.gigaweb.com/Content/Adhoc/RAH-072000-00018.html>. According to Andrew Bartels, Senior Research Analyst for Electronic Commerce, Giga Information Group, these are actually worldwide sales estimates for U.S. companies, but over 95 percent of the sales are/will be to Americans (e-mail to the author, 30 Jan. 2001). Return to text.
25. U.S. Census Bureau, “Retail Sales, by Type of Store and State: 1996 and 1997,” Table No. 1294, Statistical Abstract of the United States: 1999 (Washington: U.S. Department of Commerce, 1999) 778. Return to text.
26. Cline and Neubig 8. Return to text.
27. Austan Goolsbee and Jonathan Zittrain, “Evaluating the Costs and Benefits of Taxing Internet Commerce,” National Tax Journal, Sept 1999, University of Chicago Web <http://gsbwww.uchicago.edu/fac/austan.goolsbee/research/jzntj.pdf> 21 July 1999. Return to text.
28. Cline and Neubig 7. Return to text.
29. Cline and Neubig 7. Return to text.
30. The U.S. Advisory Commission on Intergovernmental Relations, Taxation of Interstate Mail Order Sales: 1994 Revenue Estimate. Return to text.
31. This is based on the University of Kentucky Survey Research Center data indicating that 52 percent of Kentuckians said they are either very or somewhat likely to pay the use tax. Return to text.
32. Refer to Michael T. Childress, et al., Collecting Taxes in the Cyberage, Appendix B, for a more detailed description of how these estimates are generated. Return to text.
33. The U.S. Advisory Commission on Intergovernmental Relations reports a rate of 16.5 percent in Taxation of Interstate Mail Order Sales: 1994 Revenue Estimates. Return to text.
34. Information on this initiative can be found on the Internet at <http://www.streamlinedsalestax.org>, <http://www.ncsl.org>, and <http://www.nga.org>. Return to text.